基于模型预测控制的微电网多时间尺度协调优化调度

微电网多时间尺度优化调度是消纳间歇性分布式能源的有效技术手段,针对传统基于潮流断面信息的多时间尺度优化方案易出现机组调节响应不及时、计划跟踪误差较大等问题,提出了一种基于模型预测控制(MPC)的多时间尺度协调调度方法。在日前调度阶段,综合考虑电价峰谷差、储能寿命及可再生能源随机性,建立了以系统运行成本最低为优化目标的最优经济调度模型。在日内调度阶段,为应对可再生能源日前预测误差带来的联络线功率波动,同时为确保储能满足日运行能量平衡约束,提出了一种基于MPC的日内滚动优化校正策略。采用有限时间窗内的滚动优化调度代替传统单断面优化调度,提前感知未来一段时间内的可再生能源出力及联络线计划的变化从而对机组出力进行调整,同时结合时域滚动和系统实时状态的反馈校正,更大限度地消除了微电网中不确定性因素的影响,确保了日前计划的合理性及系统运行的稳定性。以某示范微电网为例,通过算例分析验证了所提模型及算法的有效性。

Multi-time Scale Coordinated Optimal Dispatch of Microgrid Based on Model Predictive Control

Multi-time scale optimal dispatch of microgrid is an effective way to integrate intermittent distributed energy into the bulk system. In light of the problems that the conventional multi-time scale optimization scheme based on power flow is easy to appear not timely unit response, large tracking errors and so on, a multi-time scale coordinated dispatch method based on model predictive control(MPC)is proposed. During the day-ahead scheduling, an optimal economic dispatch model with the aim of minimum operation cost is developed, which takes into account peak-valley difference of electricity price, energy storage life, and the randomness of renewable energy sources(RESs). During the intra-day scheduling, an MPC-based rolling optimization correction strategy is proposed to depress power fluctuations caused by RES forecast errors at tie-line and ensure the daily energy balance of the energy storage. By using the rolling optimization scheduling of limited time window instead of the conventional single section optimization scheduling, the future changes of the renewable energy output and the tie-line plan can be perceived ahead of time so as to adjust the unit generation in advance. At the same time, by combining receding horizon and feedback correction of the real-time state of the system, the influence of uncertainty factors in the microgrid can be eliminate greatly to ensure the rationality of the day-ahead plan and the stability of the system. Finally, a microgrid demonstration project is taken as case study and simulation results verify the effectiveness of the model and algorithm proposed. This work is supported by Knowledge Innovation Important Direction Project of Chinese Academy of Sciences(No. KGCX2-EW-322)and National Natural Science Foundation of China(No. 51277170).